ABB - Use of FT-NIR for monitoring of fermentation and ... FT-NIR... · mapping of inocula quality from FT-NIR spectra of raw materials. Objective: minimise impact of flour quality

© ABB Group December 1, 2009 | Slide 1

GDCh Colloquium – 1st December 2009Use of FT-NIR for monitoring of fermentation and lyophilisation processes

Dr. Frédéric Despagne – Industry Manager Life Sciences and Fine Chemicals - ABB Analytical


PAT in biopharmaceutical manufacturing


Use of PAT in biopharmaceutical industryBioPlan Associates survey (2008, 434 respondents)

� PAT implementation plans status in biopharmaceutical manufacturing

� 58% considering PAT for new processes

� 42.6% considering PAT for existing processes

� 20% in the process of implementing PAT

� 3% have implemented PAT initiatives

0,0% 10,0%

20,0%

30,0%

40,0%

50,0%

60,0%

70,0%

80,0%

Time required to implement

Insufficient people in-house to manageimplementation

Uncertainty on how regulators will deal withinformation generated by PAT

Cost

Implementing PAT increases our risk of regulatoryevaluations due to small infractions or excursions

Regulatory

Economic

Technical


Use of PAT in biopharmaceutical industryGeneral status for biological manufacturing industry – Regulatory aspects

� Exposure to regulatory scrutiny

� Feeling that more data available means more risks of being challenged

� Fear that introduction of new analytical technologies in the product development process can be misinterpreted or misunderstood in the submission or during the pre-approval inspection.

� Current work by FDA: case study with Genentech to see if approval would be fasterwhen using PAT and decrease the overall time to validate a process so that industrywill begin adopting PAT in manufacturing.

� Lack of industry-specific guidance

� PAT currently not required by FDA for biopharmaceutical applications

� Current PAT guideline adapted to molecular entities but not to biologics � need to beextended with concepts applicable to complex biologics products.

� Authorities must recognise the fact that PAT tools will have to be more product and process-specific than with traditional drugs.


Use of PAT in biopharmaceutical industryGeneral status for biological manufacturing industry – Economic aspects

� Complex business case

� Potential benefits of implementing PAT in the long-term often mitigated by short term uncertainty and risk

� Trade-off between time and resource investments required from R&D stage and uncertain future for biological drugs in terms of clinical efficacy, regulatory approval, and commercial success

� Process adaptation required

� Actual benefits of PAT and QbD require an integrated approach to product and process development � Multiple changes needed to allow variability within a processdesign space while current manufacturing is based on fixed process parameters and end of batch QC tests

� Scheduling issues between operations if process end-point is based on real-time measurement of a CQA as opposed to pre-specified time (E.g. scheduling of purification steps after fermentation)


Use of PAT in biopharmaceutical industryGeneral status for biological manufacturing industry – Technical aspects

� Complexity of biologicals

� Proteins can be extremely heterogeneous and have complex structure � Many more CQAs than for small molecules, characterization not easy

� Fermentation media contain multiple ingredients and possible interferents

� Variability of raw materials and seed inoculum

� Broad dynamic range in product titers (µg/l to g/l)

� Biological products are heat sensitive and susceptible to microbial contamination �Need for aseptic equipments and principles from initial manufacturing steps in contrast to most conventional drugs.

� Culture yield very sensitive to process parameters variations (temperature, aeration, agitation)

Fresh Media

Dissolved O2

Temperature

Air

Offt Gas

Acid/base

pH


Use of PAT in biopharmaceutical industryGeneral status for biological manufacturing industry – Technical aspects

� Multiplicity of PAT tools

� Mass spectrometry, UV, fluorescence, on-line HPLC, chemical imaging, Raman, mid-infrared, near-infrared, …

� Identification of CPPs and CQAs � Requires tools and experts to handle analyticaldata overload

� Hardware constraints

� Availability of instruments that can self-control their operations to ease decisions by manufacturing personnel

� Most of the current process equipment is not appropriate for use with PAT � Almostimpossible to re-qualify and validate legacy equipment

� Issues associated with transfer of analytics from R&D scale to manufacturing


NIR for fermentation and cell culture applications


Use of real-time sensors for fermentation processesObjective:

� Generation of live data to understand and control fermentation and cell culture processes

Inferential sensors

� pH, temperature, oxygen, agitation, airflow, feed

� Require complex mathematical models to extract process information (“soft sensors”)

� Only surrogate markers of product quality

Spectroscopic sensors

� Direct measurement in fermentation broth of nutrients, products and by-products

� Provide status of fermentation matrix condition

� Provide direct product quality information

� Often require multivariate process models

0 5 10 15 20 25 30-0.5

0

0.5

1

1.5

2

2.5

3

3.5Soft Sensor u vs u from adaptive-filtered OD

Biomass soft sensor vs OD

Biomass via NIR vs OD


NIR spectroscopy applied to fermentationFeatures

� Fast � Allows real-time measurements

� Flexible� Fiber optics allows in-situ analysis

� Different sampling options allow to accommodate different fermentation media (clear, opaque) � Multiplex instruments

� No sample preparation required

� Accurate � Can be used as an alternative method to laboratory at-line measurements

� Information rich� Determination of chemical and physical

characteristics of samples � can be used for in-line qualitative and quantitative analysis (nutrients, products, by-products, pH)

� Can pick-up information on the complete matrix (not chemically characterized)

� For R&D, can screen a large number of fermentation process attributes

� For manufacturing can monitor CQAs

FT-NIR spectra: in-line monitoring of Saccharomyces Cerevisiae yeast growthon malt extract (ethanol production)


NIR for fermentation – Technical aspectsProcess matrix categorisation

� Classification based on bioprocess fluid and operationnal characteristics of reactor � Determines complexity of NIR modelling for key analytes1

� Selection of sampling interface

� For in-line control of manufacturing processes; in-situ probes are preferred to avoid introduction of manual sampling in sterile processes or sampling loops with potential dead zones

Increasing complexity


NIR for fermentation – Technical aspectsProbe optimisation

� Transflectance probes� Can accommodate clear (transmission) and optically dense (reflectance)

media � Useful for exploratory R&D work from lag phase to exponential and stationary phases (alternative: multiplex measurement)

� Adjustable pathlength useful for method development but issues in manufacturing (cleaning, reproducibility and stray light issues)

� Transmission probes� Useful for most fermentation processes except filamentous fermentations

(e.g. antibiotic production)

� Require aseptic design

� Reflectance probes� Useful for some filamentous fermentations

� Important constraints related to sterility regardless of probe type� Surface finish of wetted parts

� Special fiber and connector design if autoclaving required

� Special window design if CIP/SIP required (temperature gradient)


NIR for fermentation – Technical aspectsChemometrics1,2

� Spectral region optimisation (exclude C-H combination region below 4800 cm-1 and strong water band at 5100 cm-1)

� Full spectra algorithms preferred to extract complex matrix information

� Incorporate temperature fluctuations in model

� Consider time-segmented calibrations to handle fast matrix changes

� Use of spectral preprocessing

� Effect of agitation rate, gas flow rates and biomass variations on scattered light variations � baseline shifts

� Use process trajectories to detect culture contaminations

� Use on-the-fly statistical evaluation of model applicability to detect extrapolation conditions where model robustness can be challenged (e.g. F-ratios)


NIR for fermentation and cell cultureComplex raw materials identification and quality analysis

� Qualification of master cells and nutrients for fermentation step

� Monitoring of inoculum generation and prediction of inoculum quality

� Selection of the best inoculum out of a number of candidates to reduce process variability

� Qualification of enzymes and solvents for modification step

� Qualification of column packing material for chromatography and purification steps

Courtesy J. Cardoso de Menezes (IST Lisbon)


NIR spectroscopy applied to fermentation processesComplex raw materials identification and quality analysis

� Example: Classification of soybean flours from different suppliers and Kohonen network mapping of inocula quality from FT-NIR spectra of raw materials. Objective: minimiseimpact of flour quality variations (nitrogen source) on titer of final antibiotic produced by microbial fermentation of strain of mycellial micro-organisms "streptomycetesclavoligerus"3.

Predictive determination of inoculum spectra quality from flour spectra. Calibration + test set. (IST-Cipan)


Monitoring of the fermentation process

� Determination of fermentation analytes concentration (nutrients, metabolites, by-products)

� Example: monitoring of total sugars and ethanol from FT-NIR spectra during very-high-gravity corn mash fermentation (NCSU-BTEC)

NIR spectroscopy applied to fermentation processes


Monitoring of the fermentation process

� Cell density and biomass content

� Example: determination of biomass growth by correlation of FT-NIR spectra with cell density during very-high-gravity corn mash fermentation (NCSU-BTEC)



Monitoring of downstream purification process

� Monitor the elution process and fraction collection to reduce loss of API through ionic exchange column (alternative to assay that takes 10 minutes)3

� Example: Use of FT-NIR to control chromatography column gradient mixing in a protein purification process (NNE Pharmaplan)


Pool

Solvent BSolvent A

Load UV/HPLC

Waste

NIR

HPLC

PCS


Monitoring of downstream modification process

� Control of matrix condition to improve selectivity of modification process

� Example: on-line water/organic solvent ratio to optimise selectivity of fatty acid addition to a protein (NNE Pharmaplan)


Before NIR implementation


Monitoring of downstream modification process

� Control of matrix condition to improve selectivity of modification process

� Example: on-line water/organic solvent ratio to optimise selectivity of fatty acid addition to a protein (NNE Pharmaplan)

��

��

��

��

��

��

��

��


After NIR implementation


NIR for lyophilisationapplications


Use of PAT sensors for lyophilisation processesObjectives:

� Real-time monitoring of moisture content for end-point determination of freeze-drying process

� Achieve real time feedback control to achieve “right first rime” quality4

At-line methods

� Loss on drying (IR) � lack of repeatability, destructive, can not distinguish free/bonded water

� Karl-Fischer titration � destructive, slow, requires skilled operators

Some PAT sensors

� Temperature / conductivity probes inserted in vials:

� Potential crystallisation points during freeze-drying

� Sterility issues

� Non-representative information

� Mass spectroscopy (Headspace measurement on vaporised water):

� No process interference

� Provides overall process end-point information

� No direct measurement of CQA

� Requires constant leak rate � frequent leak tests required


NIR spectroscopy applied to lyophilisation processesFeatures

� Fast � Allows real-time measurements

� Flexible� Fiber optics allows in-situ analysis

� Non-contact diffuse reflectance probes allow non-invasive measurements through glass vials

� No sample preparation required

� Strong signature of water of NIR �accuracy comparable to reference methods and better precision (SEP<0.2% w/w)

� Non destructive

� Multiplex instruments provide information on several points � “mapping” of lyophiliser

� Information rich� Determination of moisture level, crystallinity

and polymorph variations

Water absorption (combination bands) in FT-NIR spectra


Possible setup

� Process FT-NIR analyser with non-contact probes inserted in lyophiliser wall

NIR spectroscopy applied to lyophilisation processes


Technical challenges – Probe-related aspects

� Availability of sampling points� Not possible to retrofit legacy equipment

� Collaboration with freeze dryer supplier required early in project

� Ideally have sampling points coupled with windows

� Positioning of sampling points� Usually only external vials accessible due to tray design with silicon oil circulation

between shelves� Monitoring only (cycle-time reduction) � OK

� Feedback control or parametric release � Requires careful validation to demonstrate representativity of measurements (drying usually not homogeneous)

� Reproducibility of probe alignment in front of vial can be challenging (automated tray loading, possibly different vial diameters depending on products) � Possibility to visually preposition retractable non-contact probe after tray loading and manually fine tune probe positioning by using XYZ 3-D stage and maximising reflectance signal after water sublimated (powder), or early in process by focusing beam on stopper.

� Specifications� Probe with aseptic design that sustain P and T conditions, including SIP cycle

temperature gradient

� Vacuum-tight flange seal



Technical challenges – Chemometrics-related aspects

� Stable measurements needed� Freeze drying cycles can reach up to several weeks � Requires stable instruments

that can run during that period with the same background

� Calibration model type� Monitoring only: qualitative trending of peak height

� Feedback control or parametric release� Matrix-specific PLS model

� Spectral peak area linear regression5

� Adaptive process software for real-time prediction (automatic switch between high moisture range model (free water) and low moisture range model (bonded water)



Conclusions


NIR for fermentation and lyophilisation processesFermentation and cell culture applications

� NIR useful at multiple stages of fermentation-related processes

� Upstream: raw materials qualification

� In-process: real-time monitoring of matrix condition and product titer

� Downstream:

� Qualitative analysis of column material + real time monitoring of the purification process

� Real-time monitoring of modification process

� Key aspects:

� Proper preliminary matrix categorisation

� Increasing availability of papers related to manufacturing implementations

Lyophilisation

� Technology potential: achieve true QbD as alternative to QC testing

� Main challenge: in-line analyser interfacing to freeze-dryer

� Expected that first advances in manufacturing will be in the field of process monitoring rather than control


Use of PAT in biopharmaceutical industryReferences

1. Scarff, M., Arnold, S.A., Harvey, L.M., and McNeil, B., Near infrared spectroscopy for bioprocessmonitoring and control: current status and future trends. Critical Reviews in Biotechnology, 2006. 26: p. 17-39.

2. Cervera, A.E., Petersen, N., Eliasson Lantz, A., Larsen, A., and Gernaey, K.V., Application of near-infrared spectroscopy for monitoring and control of cell culture and fermentation. Biotechnology Progress, 2009 (in press)

3. Lopes, J. A., Costa, P.F., Alves, T.P., Menezes, J.C., Chemometrics in bioporcess engineering: process analytical technology (PAT) applications. Chemometrics and Intelligent Laboratory Systems, 2004, 74: p. 269-275.

4. Rhoden, A., The need for PAT during freeze-drying, American Pharmaceutical Review, 2007, p. 28-31

5. Brülls, M., Folestad, S., Sparén, A., Rasmuson, A., Salomonson, J., Applying spectral peak area analysis in near-infrared spectroscopy moisture assays . Journal of Pharmaceutical and BiomedicalAnalysis, 2007, 44: p. 127–136

Acknowledgements

� Jose Cardoso de Menezes (IST Lisbon)

� Tommy Nielsen (Statens Serum Institut)

� Henry Lamb (NCSU-BTEC)

� Lucas Vann (NCSU-BTEC)

� Connie Heinze (NNE Pharmaplan / Novo Nordisk)

� Casper Leuenhagen (Novo Nordisk)


Contact: [email protected]

Web: www.abb.com/analytical

Documents

ABB - Use of FT-NIR for monitoring of fermentation and ... FT-NIR... · mapping of inocula quality from FT-NIR spectra of raw materials. Objective: minimise impact of flour quality